Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

flashing/flickering text #223

Closed
ubemotho opened this issue Mar 22, 2024 · 5 comments
Closed

flashing/flickering text #223

ubemotho opened this issue Mar 22, 2024 · 5 comments
Labels
invalid This doesn't seem right

Comments

@ubemotho
Copy link

ubemotho commented Mar 22, 2024

when enable highlight_words text is flicker like in this example openai/whisper#1072

The.Hacker.Quarantine.Short.Film.1080p_25fps_H264-128kbit_AAC._burn-in_1920x1080_x264.mp4

i used this for example https://www.youtube.com/watch?v=ErRr-vA7_-U

i dont know if this a problem from whisperx or faster-whisper implementation
but it should be fixed in whisper in this commit openai/whisper#1087

@Purfview
Copy link
Owner

Can you post your whole command used?

@ubemotho
Copy link
Author

--compute_type float16 --highlight_words true

-----------------------------------------------------------------------------
Date: 03/22/2024 
SE: 4.0.4.0 - Microsoft Windows NT 10.0.19044.0 - 64-bit
Message: C:\Users\X\AppData\Roaming\Subtitle Edit\Whisper\Purfview-Whisper-Faster\whisper-faster.exe --language en --model "tiny.en" --compute_type float16  --highlight_words true "C:\Users\X\AppData\Local\Temp\a80d86ec-3c06-404c-88a7-2bbcdb1e3d44.wav"

-----------------------------------------------------------------------------
Date: 03/22/2024 
SE: 4.0.4.0 - Microsoft Windows NT 10.0.19044.0 - 64-bit
Message: Calling whisper (Purfview's Faster-Whisper) with : C:\Users\X\AppData\Roaming\Subtitle Edit\Whisper\Purfview-Whisper-Faster\whisper-faster.exe --language en --model "tiny.en" --compute_type float16  --highlight_words true "C:\Users\X\AppData\Local\Temp\a80d86ec-3c06-404c-88a7-2bbcdb1e3d44.wav"
Standalone Faster-Whisper r186.1 running on: CUDA
Starting transcription on: C:\Users\X\AppData\Local\Temp\a80d86ec-3c06-404c-88a7-2bbcdb1e3d44.wav
[00:15.270 --> 00:37.620]  Activate the intelligent system. Access granted. Good evening, Bennett. How are you? Being in quarantine isn't so great. Thank you for asking. I am so lucky to be a bot. Shut up. Just give me information on the current coronavirus situation.
[00:37.620 --> 01:05.940]  There are nearly 2 million confirmed cases of coronavirus worldwide. The number of infections with a new coronavirus continues to rise. Is there a vaccine on the COVID-19? There is currently no vaccine but there is a patient that has recovered from COVID-19 that has proven to have the most effective antibody against the virus. Do we have more information on the patient? Information classified. Shit. I guess I have to do it myself.
[01:10.900 --> 01:43.160]  Access granted. Yes. Got her name is Sena Savana. She is 26 year old and is a kindergarten teacher and is currently kept held in a Swiss biotech firm. Oh my god. She's the back scene. Shit. You have 10 seconds left. Starting count down now. 8, 7, 6, 4, 5, 4, 3, 2, 1.
[01:43.160 --> 01:55.340]  Down low. You're too late.
Transcription speed: 47.02 audio seconds/s

Operation finished in: 4 seconds

Subtitles are written to 'C:\Users\X\AppData\Roaming\Subtitle Edit\Whisper\Purfview-Whisper-Faster' directory.
Calling whisper Purfview's Faster-Whisper done in 00:00:08.8228101
Loading result from C:\Users\X\AppData\Roaming\Subtitle Edit\Whisper\Purfview-Whisper-Faster\a80d86ec-3c06-404c-88a7-2bbcdb1e3d44.srt

@Purfview
Copy link
Owner

I think that's Subtitle Edit issue. Disable SE's post-processing!

Ping to @niksedk

@Purfview Purfview added the invalid This doesn't seem right label Mar 22, 2024
@ubemotho
Copy link
Author

ubemotho commented Mar 22, 2024

I think that's Subtitle Edit issue. Disable SE's post-processing!

Ping to @niksedk

oh youre right. i am new to subtitle edit i forgot about post-process. sorry for your time and thank you. i added a example for better understanding

here is more info for SE guys
this is SE output:

1
00:00:15,270 --> 00:00:15,578
<u>Activate</u> the intelligent system. Access granted. Good evening,
Bennett. How are you? Being in quarantine isn't so great.

2
00:00:15,590 --> 00:00:15,910
Thank you for asking. I am so lucky to be a bot. Shut up.
Just give me information on the current coronavirus situation.

3
00:00:15,911 --> 00:00:16,129
Activate <u>the</u> intelligent system. Access granted. Good evening,
Bennett. How are you? Being in quarantine isn't so great.

4
00:00:16,141 --> 00:00:16,370
Thank you for asking. I am so lucky to be a bot. Shut up.
Just give me information on the current coronavirus situation.

5
00:00:16,371 --> 00:00:16,439
Activate the <u>intelligent</u> system. Access granted. Good evening,
Bennett. How are you? Being in quarantine isn't so great.

this is faster-whisper.exe when i run from cli with these commands: .\whisper-faster.exe '.\The Hacker (Quarantine Short Film) (1080p_25fps_H264-128kbit_AAC).mp4' -l English -m tiny -o source --compute_type float16 --highlight_words true

Details

1
00:00:14,850 --> 00:00:15,410
<u>I</u> activate the intelligent system.

2
00:00:15,410 --> 00:00:15,970
I <u>activate</u> the intelligent system.

3
00:00:15,970 --> 00:00:16,310
I activate <u>the</u> intelligent system.

4
00:00:16,310 --> 00:00:16,670
I activate the <u>intelligent</u> system.

5
00:00:16,670 --> 00:00:17,450
I activate the intelligent <u>system.</u>

niksedk added a commit to SubtitleEdit/subtitleedit that referenced this issue Mar 22, 2024
@niksedk
Copy link

niksedk commented Mar 22, 2024

Turn off SE's "Post-processing" when using the ---highlight_words true parameter

SE will do this automatically in next SE update

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
invalid This doesn't seem right
Projects
None yet
Development

No branches or pull requests

3 participants